© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW
    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    2. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    3. Gemma-2-9B
    4. Residual Stream - 131k
    5. 24-GEMMASCOPE-RES-131K
    6. 340
    Prev
    Next
    INDEX
    Explanations

    instances of the word "justify" and its variations, indicating focus on justification and rationalization themes

    oai_token-act-pair · gpt-4o-miniTriggered by @bot

    justification for actions or statements

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-9b-pt-res/layer_24/width_131k/average_l0_97
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    131,072
    Data Type
    float32
    Hook Name
    blocks.24.hook_resid_post
    Hook Layer
    24
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    Ball
    -0.46
     ball
    -0.44
    EB
    -0.41
     Velasco
    -0.41
    <em>
    -0.41
    catraz
    -0.41
     aufnehmen
    -0.40
     Batam
    -0.40
     Ball
    -0.40
    wc
    -0.40
    POSITIVE LOGITS
     Justification
    1.04
     justifies
    1.00
     Jus
    0.95
     justifying
    0.94
     justification
    0.94
     gius
    0.90
     justified
    0.89
    Justification
    0.88
     justify
    0.87
    justified
    0.85
    Activations Density 0.014%

    No Known Activations